1. In Chan Song, Ph.D. Seoul National University Hospital
Propeller MRI
In Chan Song, Ph.D.
Seoul National University Hospital

2. Contents: Propeller sequence
(Periodically Rotated Overlapping Parallel Lines with
Enhanced Reconstruction)
Motion artifact
Theoretical basis
Applications

3. Motion Cause  Periodic: cardiac motion, respiration, blood flow
Sporadic: irritable patients’ motion
Translation, rotation, through-plane 
Artifact in MRI
blurring and ghosting
Cause
Longer encoding step

4. Scan time=
TR x matrix x Average
 Long scan time

5. MR image reconstruction
under the assumption of object’s motion-free condition
during whole k space coverage

6. Motion artifacts -Most ubiquitous and noticeable artifacts in MRI
due to voluntary and involuntary movement, and flow (blood, CSF)
-Mostly occur along the phase encode direction, since adjacent lines of phase-encoded protons are separated by a TR interval that can last 3,000 msec or longer
-Slight motion can cause a change in the recorded phase variation across the FOV throughout the MR acquisition sequence

7. Motion artifact: ghost and blurring

8. Solution for motion compensation
-Navigator echo usage to estimate the motion or motion related
phase from extra collected data
-Cardiac and respiratory gating
-Respiratory ordering of the phase encoding projections based on
location in respiratory cycle
-Signal averaging to reduce artifacts of random motion
-Short TE spin echo sequences (limited to spin density,
T1-weighted scans). Long TE scans (T2 weighting) are
more susceptible to motion

10. Motion (abrupt)  phase error  position error
Solution
Phase information
Navigation
 Motion correction by phase information

11. Key ideas in propeller sequence
K space: partial covering for whole image
Motion detection: blade usage
Correction: FFT properties’ usage

12. Diagram of the PROPELLER collection reconstruction process for motion corrected MRI.

13. Data acquisition
Propeller filling
Rectangular filling kx ky

14. Phase Correct
Redundant data must agree, remove phase from each blade image
Imperfect gradient balancing,
Eddy current effect:
 echo center shift

16. James G. Pipe

17. Windowing
Before
After
Phase correction

18. Bulk Transformation Correction
Fourier transform correspondence
Image space  k space
Translation  Phase roll
Rotation  Rotation
Separate estimation of rotation and translation

19. rotate imagerotate data
Fourier Transform Properties
rotate imagerotate data

20. Rotation correction (magnitude image)
Reference
(only inner circle)
Magnitude of the average of strips
Rotation
(only inner circle)
Correlation

21. Blade by blade operation
Rotation at maximum correlation
 Correction

22. Fourier Transform Properties
shift image  phase roll across data
b is blade image, r is reference image

23. max at Dx

24. Translation
Complex average k-space data
Reference
(only inner circle)
Complex of the average of strips
Multiplication
Inverse FT (maximum)

25. Blade by blade operation
Translation at maximum correlation
 Correction

26. Blade Correlation
throw out bad – or difficult to interpolate - data

27. Through-plane motion
:low weighting coeff.

28. Reconstruction (FFT)
non-Cartesian sampling
requires gridding  convolution Kx Ky

29. w/motion correction

30. no correction
correlation correction only
motion correction only
full corrections

31. Artifact reduction due to head motion
T2-FSE
T2-Propeller
T2-Propeller(corrected)

32. DWI-EPI
B=1000s/mm2
DWI-Propeller
(FSE)
James G Pipe, 2002

33. DWI (b=700s/mm2)
a. EPI
(TR/TE/avg=2700/113/15)
b. Propeller EPI
(TR/TE/blade=1600/70/26)
Wang FN, 2005

34. Useful application in propeller sequence
Motion- or Bo-inhomogeneities – insensitive
Irritable patient
Diffusion weighted image
Limitations in propeller sequence
Redundant acquisition 
Long scan time:
High SAR: problem in higher field MR system
Solutions 
Undersampling (Konstantinos Arfanakis, 2005)
Parallel imaging
Turbopropeller (James G Pipe, 2006)
Propeller EPI

35. Propeller sequence
Low sensitivity to image artifacts,
Bo inhomogeneity and motion
T2-, Diffusion-weighted images
(High SNR, low geometric distortion, low SAR)

36. References 1. Pipe J, MRM 42(5): 963-62,1999.
2. Pipe J, et al., MRM 47(1): 42-53,2002
3. Wu Y, Field AS, Alexander AL. ISMRM, Toronto, Canada,
4. Roberts TP, Haider M. ISMRM, Kyoto, Japan,
5. Sussman MS, White LM, Roberts TP. ISMRM, Kyoto, Japan,
6. Pipe J and Zwart N. Magn Reson Med 55:380–385, 2006.
7. Cheryaukaa AB, et al. Magnetic Resonance Imaging 22: , 2004